7 Atomic Force Microscopy Studies of Semicrystalline ... - Springer Link

7 Atomic Force Microscopy Studies of Semicrystalline Polymers at Variable Temperature Dimitri A. Ivanov1, and Sergei N. Magonov2 1

2

Laboratoire de Physique des Polymères, CP 223, Université Libre de Bruxelles, Boulevard du Triomphe, B-1050 Brussels, Belgium, Tel.: +32 2 650 57 59, fax +32 2 650 56 75, [email protected] Digital Instruments/Veeco Metrology Group, 112 Robin Hill Rd., Santa Barbara, CA 93117, U.S.A.

Abstract. The capabilities of Atomic Force Microscopy, AFM, in studies of polymer phase transitions and, more generally, the semicrystalline polymer morphology are described. It is shown how variable temperature AFM can provide unique information on the organization of semicrystalline polymers at the nanometer scale and its evolution in the course of crystallization. The practical examples selected for this work illustrate applications of AFM to structural studies of homopolymers and polymer blends crystallized in the bulk, in thin films and in solutions. The investigated systems include solution grown single crystals of polyethylene, cold-crystallized poly(ether ether ketone), as well as melt-crystallized poly(ethylene terephthalate), poly(trimethylene terephthalate), syndiotactic polystyrene, poly(ε-caprolactone), isotactic and syndiotactic polypropylene. The issues related to AFM image analysis and its quantitative comparison with the results of complementary techniques such as small-angle X-ray scattering (SAXS) are addressed. More specifically, it is shown how AFM can provide statistically meaningful parameters for the semicrystalline structure and an accurate choice of a structural model for the interpretation of SAXS data.

7.1

Introduction

During the last 10–15 years, the development of scanning probe microscopy and its applications to studies of different materials has been in focus of many researchers. The invention of scanning tunneling microscopy, STM, in 1982 was the first milestone [1], which opened direct access to atomic-scale surface structures. In STM, a measurement of the current flowing between a sharp metallic tip and a conducting or semiconducting surface, is used for atomic-scale imaging of surface structures. The ability to get atomic resolution on samples in ambient conditions and even under liquid substantially enhances the capabilities of STM. Yet, a severe limitation related to the requirement of sample conductivity motivated researchers to seek for a more universal tip-sample interaction, which can be applied for high-resolution surface imaging. This demand has been satisfied

Corresponding author.

G. Reiter, J.-U. Sommer (Eds.): LNP 606, pp. 98–130, 2003. c Springer-Verlag Berlin Heidelberg 2003

7

Semicrystalline Polymers at Variable Temperature

99

with the appearance of atomic force microscopy, AFM [2]. Attractive and repulsive forces acting between a sharp probe and the sample surface are employed for surface imaging with this technique. Several innovations such as optical level detection [3], use of microfabricated probes and new imaging modes [4] made AFM a well-established surface characterization tool applicable to a very broad spectrum of materials ranging from solid inorganic crystals to partially or fully disordered organic liquid-like systems. Nowadays, AFM is used in many laboratories specialized in various fields of materials science. Such rapid expansion of this technique can be explained by further important instrumentation developments significantly extending the capabilities of AFM and related methods beyond simple topography measurements, which now also include measurements of local mechanical, adhesive, thermal, as well as magnetic and electric properties. One of such very recent instrumentation developments in this field consists in the heating stage accessory designed for in situ work between room temperature and approx. 250 ◦ C [5]. Although the advantages of in situ high-temperature AFM are evident, especially for applications to organic materials, such studies have been limited so far by the absence of appropriate thermal accessories [6]. It is, therefore, expected that numerous AFM results on the thermal behavior of organic surfaces will soon be generated using this enhanced technical feature. Of special interest for the present work are studies of soft materials, such as synthetic and natural polymers [7]. The nondestructive character of AFM presents an important advantage in studies of such systems, as they can be destroyed during preparation or observation with classical techniques such as Transmission Electron Microscopy, TEM. It should be noted that studies of surfaces with AFM do not require any special sample preparation such as metal or carbon deposition or micro-sectioning or staining. Therefore, AFM can provide a great wealth of information on the structure of such systems, without causing any damage to the sample. In particular, the evolution of the sample structure at the nanometer scale can be studied in situ using appropriate conditions, e.g., under controlled atmosphere or in liquids. The softness of materials to be studied could be even advantageous for this technique since, in some instances, it could help extracting structural information from AFM observations at different levels of the tip-sample interaction. For example, low-force imaging is optimal for the correct determination of surface profiles of such samples, whereas imaging at elevated forces can be useful for surface compositional analysis and for recognition of individual components in heterogeneous polymer materials [8] or biological systems [9]. The objective of the present work is to give a brief description of the capabilities of AFM in studies of semicrystalline polymer morphology and phase transitions. The structure of this chapter is the following: At first, some technical issues related to the tip-sample interaction in tapping mode, image resolution and requirements as to sample preparation will be shortly addressed. In the main part of the review, several examples of morphological analysis of polymer surfaces including homopolymers and polymer blends crystallized in the bulk, in thin films and solutions will be presented to illustrate the unique informa-

100

Dimitri A. Ivanov and Sergei N. Magonov

tion AFM can deliver. In particular, the structural data provided by AFM will be quantitatively compared with the results of conventional techniques such as small-angle X-ray scattering, SAXS. It will be shown, for example, how the direct space information obtained from AFM images can be used to correctly model the SAXS curves.

7.2

Studies of Polymers in Tapping Mode

Description of any scanning probe microscopy technique logically includes a physical model for the tip-sample interaction and different technical issues related to the detection of this interaction and scanning operation of the probe over the sample surface. In the contact mode initially introduced for AFM a sharp tip fixed at the extremity of the cantilever comes into direct contact with the sample, which causes the cantilever deflection. A laser beam reflected from the backside of the cantilever is used for measurements of its deflection. A microscope mode of AFM is realized when the tip being scanned over the sample surface is controlled with a feedback mechanism, which adjusts the vertical sample position to keep the tip-sample interaction force constant [10]. This operation results in a topography image, which presents a three-dimensional surface profile. Surface structures with characteristic shapes and dimensions are recognized in such images given in color or grayscale codes, where brighter zones typically represent elevated surface features. Changes of the cantilever’s parameters such as deflection, frequency or phase are usually recorded during scanning and they are also collected in AFM images. Such images might reflect not only topographic features but also local sample properties (friction, adhesion, stiffness, etc.). Therefore, in its applications to polymers, AFM combines the possibilities of probing local material properties and surface imaging. It is worth noting, however, that visualization of surface structures is the major AFM application, in which AFM complements optical and electron microscopy. As will be illustrated below, AFM is also invaluable in combination with diffraction techniques for justification of the structural models used to analyze the reciprocal space data. 7.2.1

Tip-Sample Forces

Already in early AFM studies of polymer samples, it had been noticed that contact mode operation might cause modifications or even damage of the surface of soft materials. In this mode the probe scans the surface while remaining in permanent contact with it. In this case, the cantilever feels not only normal forces, which are employed for feedback control, but also the lateral tip-sample forces. The latter could be used for studies of friction, yet the same forces generate shearing deformation that leads to modification and damage of soft materials. This circumstance has limited AFM studies of polymer materials until the appearance of a new oscillatory mode, known as tapping mode [4], which was introduced to avoid the described effect of lateral forces. In tapping mode, the probe oscillates close to its resonant frequency and the variation of the cantilever amplitude due

7


101

to tip-sample force interactions is employed for surface profiling. In this mode, the lateral tip-sample forces are largely eliminated, however, consideration of the tip-force effects is still an important issue governing the image resolution and AFM capabilities for exploring polymer samples. It is worth noting that in contrast to the contact AFM mode, in which the cantilever deflection is directly related to the tip-sample force, the oscillatory nature of tapping mode makes the description of the tip-force interactions more involved. The changes of the cantilever amplitude do not describe the probe interaction with the sample to a full extent. Changes of resonant frequency and phase of the oscillating probe are most sensitive to the tip-sample forces and can discriminate between the overall attractive and repulsive force regimes. Therefore, AFM phase images, which express local differences between phases of a free-oscillating probe and the one interacting with the sample, became useful for many AFM applications, e.g. for compositional mapping of heterogeneous polymer systems [11]. A general correlation between the phase contrast and the energy dissipated by the probe interacting with the sample per oscillation cycle has been discussed [12]. Unfortunately, this does not help to explicitly correlate the phase changes in terms of differences in specific material properties such as adhesion, stiffness or viscoelasticity. A detailed description of the tip-sample interaction in the tapping mode experiment on a polymer sample is extremely complex. It should take into account a large number of phenomena and factors including the cantilever vibration amplitude, cantilever stiffness, sample stiffness and adhesion, tip shape, penetration of the tip into the sample, viscoelasticity (which is, most likely, nonlinear under tapping mode conditions). Although numerous experimental and theoretical efforts are directed towards understanding these issues, imaging of model samples with components having different physical properties (stiffness, density, viscosity, adhesion, etc.) might be a good starting point helping to rationalize the tip-sample interactions and their correlation with the oscillation parameters (amplitude, frequency, phase, etc.). In one of such attempts [8], the surface of a blend of two polyethylene materials with different densities (0.92 and 0.86 g/cm3 ) was examined at different free oscillation amplitudes and setpoints. It was found that the difference in the component’s density resulted in an apparent height variation caused by depression of the tip into the low–density component. This effect is most pronounced when the initial cantilever amplitude is sufficiently large (80−100 nm) and the set-point amplitude is around 40−50% of the initial amplitude. Such imaging at elevated forces, which is also known as hard tapping, is usefull when the compositional mapping of multi–component systems is desirable [13]. In the described case, the apparent contrast in the height image reveals the presence of components with different density and stiffness (the mechanical moduli differ approx. by a factor of 1000). In studies of other heterogeneous materials, in which the difference in the properties of components is not so drastic, the contrast in phase images is used for the identification of components. Contrary to hard tapping, imaging at a low-level of tip-sample forces (light tapping) is realized when the set-point amplitude is close to the amplitude of the

102


free-oscillating probe (A0 ). The tip-sample forces are even smaller when A0 is small. Unfortunately, the latter is not always achievable because of capillary or adhesive force. Practical use of light tapping is obvious: such imaging conditions are applied when high-resolution profiling of the top surface is needed for example for imaging of single macromolecules deposited on the substrate or for correct estimate of surface roughness of technologically important surfaces. When a top surface layer is in the rubbery state, imaging in light tapping reveals the top layer morphology. However, with the increase of the force the tip will penetrate the top layer and reveal the morphology of the sub-surface material. This effect was observed in studies of block copolymer films [14] and polyethylene films coated with low-molecular weight material [8]. In imaging of single dendrimer macromolecules, which are seen as 5-nm spheres, a transition from light to hard tapping revealed a harder core of these spheres related to a higher density in the macromolecule center [15]. 7.2.2

Image Resolution

Image resolution is an important characteristic of AFM. In contact mode, the images of flat surfaces of organic and inorganic crystals can correctly reproduce the periodic atomic structures of the surface. However, the image resolution of real polymer samples is a more complex issue, which is determined not only by the tip-sample contact area, but also by the sample nature and roughness. For example, the width of isolated macromolecules deposited on a flat surface is overestimated when measured from AFM images taken even in light tapping. This is due to the effect of dilation of a real profile of the extended macromolecules √ by the tip geometry, which gives rise to an artificial width increase of about 2Rh, where R is the tip radius and h the height of the step. For the same reason, the estimates of the macromolecule width become more precise when performed from images of compact macromolecular stacks. AFM measurements of the contour length of such objects are more accurate, and they could be even applied to study molecular weight distribution of polymers [16,17]. Object orientation is also of importance for determination of its dimensions from AFM images. For example, to get a correct estimate of the interlamellar distance in a lamellar stack (also known as long period, LB , determined by SAXS), one should select the stacks with the smallest intercrystalline spacing. These objects, given the sufficient image statistics, should be close enough to edge-on orientation. It is worth noting that by varying the tip-sample force during imaging of lamellar structures one can clarify the thickness of the most compact, crystalline regions (denoted here as Lc ) and also the size of a less compact amorphous exterior of crystals (La = LB − Lc ) [8]. Since, as was mentioned before, the morphological details pertinent to the bulk polymer structure are often hidden underneath a featureless top layer of amorphous material, the crystalline lamellae are visualized with higher contrast in the images obtained in hard tapping.

7

7.2.3


103

Sample Preparation

Sample preparation is one of the crucial steps of microscopic studies with AFM. It is clear that, for example, the sample surface roughness can present a practical limitation for AFM. Since most piezo actuators used in atomic force microscopes have only a few microns vertical range, AFM studies are restricted to surfaces with corrugations not exceeding this value. Also imaging of fine features with steep slopes is quite difficult because of imperfections of the feedback and crosstalk between motion of vertical and lateral actuators. Therefore, flat surfaces are most preferable for examination of fine morphological features of polymer samples. Many commercial films having surface roughness of a few hundreds of nanometers can be examined without any surface preparation. Films prepared by casting polymer solutions on flat substrates or by hot pressing between smooth plates such as mica sheets are much smoother, and they are well suited for high-resolution AFM studies. In preparation by hot pressing, the polymer film is separated from the molding plates at room temperature and the opened surface is examined with a microscope. Films of semicrystalline polymers with Tg higher than room temperature prepared by this method might be quite flat but constrained upon cooling in-between the molding plates. In such a case, thermal annealing of free-standing films close to Tg will be essential for obtaining a sample with a more equilibrated surface structure. However such annealing is usually accompanied by surface roughening. AFM is primarily a surface characterization technique, and access to the bulk polymer morphology requires surface etching, ultramicrotomy or a combination of both techniques. Chemical or plasma etching removes a top surface layer whereby amorphous material is usually etched faster than crystalline structures. Etching recipes are documented for many semicrystalline polymers [18] and can be successfully applied for AFM sample preparation. Ultramicrotomy conventionally used for preparation of thin (30–100 nm) sections of polymers for TEM studies is also applicable for AFM studies. It should be noted that ultramicrotoming of polymers with Tg above room temperature is usually done at room temperature, whereas polyolefins and other polymers with Tg below room temperature require the use of a cryo-stage at temperatures 10–20 degrees lower than Tg . Roughness of polymer surfaces prepared by sectioning with a diamond knife, which is by far more appropriate for this procedure than a glass knife, is typically of several hundreds of nanometers. However, the quality of cuts may be subject to variations due to the material properties, and therefore the lamellar structure of semicrystalline polymers is not always resolved in AFM images of microtomed surfaces.

7.3

Practical Examples of Imaging of Semi-crystalline Polymers at Variable Temperature

The sensitivity of AFM to local material properties opens access to studies of polymer crystallization. In this section, several examples illustrating capabilities

104


of AFM in such studies will be given. It should also be noted that the morphological observations of semicrystalline polymers reported in this work cannot be fully interpreted in the frame of the classical one-dimensional picture of the semicrystalline structure composed at its elementary level of interleaved sheets of amorphous and crystalline regions. In this respect, AFM data providing evidence for the existence of other types of organization in semicrystalline polymers, e.g., the much-debated granular morphology, will deserve special attention. The section will start with the most studied group of flexible chain polymers showing some examples where AFM can bring additional information with respect to conventional techniques of morphological analysis. In the second part, the morphology of the somewhat less studied family of semirigid chain semicrystalline polymers will be addressed. 7.3.1 7.3.1.1

Flexible Chain Polymers Solution Grown Single Crystals of Polyethylene, PE

Generally, the crystallization of polymers proceeds by multiple folding of chain macromolecules to form two-dimensional crystalline lamellae. This process is well defined in dilute polymer solutions where only few chain entanglements are present. Individual lozenge-shaped lamellae of PE grown in solutions are probably the best-explored polymer crystals [18,19]. These crystals, which form initially as hollow pyramids, collapse into two-dimensional platelets while being collected on the substrate. The length of a straight segment of polymer chains defines the lamellar thickness, which is usually in the 10–12 nm range. Prior to AFM, TEM in combination with X-ray diffraction has been applied for the examination of single PE crystals, and the basic structural features of these objects have been explored in depth. A combination of TEM imaging and electron diffraction made it possible to observe the sectorization and determine the polymer chain orientation inside individual crystal sectors. However, many questions related to the organization of single crystals remain open. One of such questions concerns the structure of the lamellar surface, which is presumably formed by chain folding according to the adjacent reentry model. There are also alternative models of this structure with different degrees of order. With the advent of AFM, researchers were motivated to use its advantages over TEM such as precision of height measurements and high-resolution imaging of top surface features in studies of polymer single crystals. Indeed, single crystals of PE have been examined in AFM images with great structural detail such as sectors, screw dislocations and overgrowth [20]. More recently, 8–10 nm size grains were observed in low-force images of the crystal surface acquired in tapping mode, whereas traces of lattices relevant to crystallographic register of polymer chains were detected in images obtained in contact mode with relatively high forces [21]. AFM imaging of single PE crystals has also been performed at elevated temperatures at which lamellar thickening usually takes place [22]. It should be noted that this process is caused by unfolding of individual chains from a kinetically favorable folded state to an energetically favorable extended-chain

7


105

Fig. 7.1. Height images (14 × 14 µm2 ) of dry single crystals of polyethylene measured at room temperature (A) and 115 ◦ C after 1.5 hr annealing (B). (C) Height histograms corresponding to AFM images of dry single crystals of PE, which show the evolution of lamellar thickness after annealing at different temperatures. (D)–(E) Height images (10 × 10 µm2 ) of a not completely dry PE single crystal. Images obtained at room temperature before and after annealing at 105 ◦ C, respectively. (F) Part of the PE single crystal given in Fig. 7.1E showing edge-on lamellae (4 × 4 µm2 scan).

conformation [23]. AFM observations of crystal annealing reveal formation of holes surrounded by thickened regions, which is consistent with the earlier TEM data [19]. In high temperature AFM images shown in Fig. 7.1A–B it can be seen that holes appear simultaneously with thickening of the adjacent locations. Thickening proceeds gradually after a step-wise change at 115 ◦ C, as reflected in the height histograms in Fig. 7.1C. Thus, the sensitivity of AFM to height measurements makes it possible to monitor chain unfolding in situ with exceptional precision. A detailed comparison of the thickening mechanisms taking place in single crystals of PE and regular linear alkanes of different length [24] can be useful in understanding the role played in these processes by the crystal/amorphous interphase and the polymer nature of the reorganizing species 1 . As shown in the early TEM studies [19], annealing of single PE crystals containing traces of solvent (xylene) proceeds differently, and the thickening can be accompanied by reorientation of chains, which become aligned along the crystal surface. 1

see also Chap. 6 for more details.

106


These observations are in good agreement with the AFM data presented in Fig. 7.1D–F. Indeed, upon annealing of a not completely dry PE crystal, the formation of holes (some of them are barely seen in Fig. 7.1E) takes place simultaneously with polymer chain reorientation as evidenced in Fig. 7.1F by stacks of the edge-on lamellae [25]. In this case, AFM offers the possibility to study the dynamics of chain reorientation in single PE crystals, which can provide deeper insight into the elementary steps of the chain unfolding [26]. 7.3.1.2 Melt-Crystallized Polyolefins: HDPE, LDPE, Isotactic and Syndiotactic Polypropylene Polymer crystallization from the melt is a much more complex phenomenon compared to solution crystallization, and its mechanisms are far from being completely understood. It is documented that crystal growth from the melt leads to formation of characteristic morphological patterns, among which spherulitic morphology is probably the best known. AFM images of various polyethylene and polypropylene samples can illustrate different aspects of crystalline morphology and its dependence on regularity of individual polymer chain structure. It is expected, for example, that extensive folding of macromolecular chains into lamellae and compact arrangement of lamellae inside the spherulites will be characteristic of high-density polyethylene, HDPE. This is, indeed, what is confirmed by AFM images obtained on microtomed and etched surfaces of this polymer shown in Fig. 7.2B–H. The semicrystalline morphology of the sample reveals banded spherulites, as observed with polarized optical microscopy (Fig. 7.2A). The optical contrast reflects variations in birefringence [27], which is the result of different local orientations of polymer chains. AFM images of the same surface (Fig. 7.2B) also show banded patterns, with the height contrast corresponding to the surface corrugation of the etched sample [28]. At higher magnifications (Fig. 7.2C–D), one can notice that elevated regions are mainly composed of stacks of relatively flat-lying lamellae. By contrast, the structures visible in the valleys between the bright bands can be identified with edges of the lamellae oriented perpendicular to the sample surface. Of special concern are granular structures, which are better resolved in the phase image in Fig. 7.2D. The granules can be seen on lamellar edges and also on the surface of flat-lying lamellae. During the past 20 years, there have been a number of electron microscopy observations of similar structures in semicrystalline polymers [29,30]. However, the question of whether these structures are real or originate from sample preparation, which typically includes etching or staining, has not yet been clarified. Additional evidence for the existence of such structures has been obtained in AFM studies of polymer surfaces, which were not subject to etching or staining [28], [31]. For example, a large-scale image of the surface of the melt-crystallized HDPE reveals spherulites with a relatively smooth surface (Fig. 7.2E). The spherulitic bands are formed by crystals oriented in tangential and radial directions (Fig. 7.2F–G), where, at higher magnification, individual fibrils clearly exhibit a grainy surface (Fig. 7.2H).

7


107

Fig. 7.2. (A) Optical micrograph of a microtomed sample of HDPE after permanganate etching, micrograph size is 35 × 35 µm2 . (B) AFM height image of the same surface (35 × 35 µm2 scan). (C)–(D) Height and phase images (1.5 × 1.5 µm2 ) of a part of the banded HDPE spherulite shown in (B). (E) Height image of a banded HDPE spherulite recorded on the surface prepared by hot molding between mica sheets (35 × 35 µm2 scan). (F–G) Height and phase images (6 × 6 µm2 ) of a part of the spherulite shown in (E). (H) High-resolution height image of the same spherulite as in (e) (700 × 700nm2 scan).

In contrast to HDPE, highly branched chains of low-density polyethylene, LDPE, are less prone to multiple folding with adjacent chain reentry. At the scale of several microns, melt-crystallized samples of LDPE reveal ill-defined spherulites, also known as sheaf-like structures (Fig. 7.3A–B). However, at highermagnification, similarly to HDPE, the surface of these sheafs appears to be decorated with numerous grains, especially when imaging is performed in light tapping (Fig. 7.3C). With the increase of the tip-sample force, the image taken at the same location shows only slightly curved fibrils (Fig. 7.3D), which is similar to typical morphological patterns observed with TEM (Fig. 7.3E). The described changes are fully reversible, and grainy morphology is restored when the tip-sample force is lowered again. Therefore we speculate that the fibrils consist of a more rigid core and a rubbery-like exterior [28], where crystalline stems form the core, while less ordered or amorphous material with Tg far below room temperature constitutes the exterior. Although an ideal lamella can be considered as a two-dimensional platelet, one can easily imagine its degeneration into a fibril (i.e., “one-dimensional” object) or a grain (Fig. 7.3F) in the presence of structural defects or geometrical constraints (cf. Fig. 7.1F). In this case, the observed grains would reflect the micro-blocks of the underlying crystalline structure. It is however, not clear whether the observed granular morphology is characteristic of the polymer organization in the bulk or only at the surface. In many instances, the grainy appearance of the polyolefin polymer surfaces can be

108


Fig. 7.3. (A)–(B) Large-scale height images of the surface of LDPE hot-molded between mica sheets, 60 × 60 and 12 × 12 µm2 scans, respectively. (C)–(D) Height images (680 × 680 nm2 ) of the same part of the sheaf-like spherulite shown in (B) recorded in light (C) and hard tapping (D). (E) Typical TEM image of the lamellar structure of PE (640 × 640 nm2 size micrograph). (F) Schematic drawing showing the proposed structure of fibrils and lamellae observed in AFM images of melt-crystallized polymers.

7


109

smeared out by increasing the tip-sample force, which probably indicates that this structure is more related to the surface organization. Fibrils and grains can be found not only on surfaces of melt-crystallized bulky polymer samples, but also in thin and ultra-thin films, which is demonstrated in Figs. 7.4A–E. The first two images show morphology of an ultra-thin layer of ultra-high molecular weight, UHMW, PE formed by dipping a piece of silicon wafer into a dilute polymer solution in xylene. Concentric patterns of crystalline material result from the interplay between dewetting, crystallization and solvent evaporation. A flaton lying lamella with grainy surface seen in the center of Fig. 7.4B is surrounded by short fibrils with a block substructure, which are attached to each other. The width of the fibrils is between 10 and 15 nm and the height of the flat-on lamella is approx. 20 nm, which is in the range of a typical lamellar thickness.

Fig. 7.4. (A)–(B) Height images of ultrathin UHMW PE film obtained by dipping a Si wafer into diluted xylene solution at high temperature, 5 × 5 µm2 and 600 × 600nm2 scans.

It is interesting to note that the presence of granular morphology is not necessarily related to the lamellar organization of the semicrystalline structure. Thus, in contrast to UHMW PE, a metallocene-based polyethylene sample with high concentration of octene branches does not reveal well-defined lamellar structures (Figs. 7.5A–B). The image, which was recorded in light tapping, exhibits mostly grainy surface features. When imaged in hard tapping (Fig. 7.5B), the same location looks different and short fibrils are seen everywhere. The transition between the two morphologies is fully reversible and seems consistent with the fibrillar structure schematically depicted in Fig. 7.3F. It is instructive to compare the semicrystalline morphology and the structure of the top surface layer of PE with that of another important member of the polyolefin family, polypropylene. Morphologies of a thin film of fractionated syndiotactic polypropylene, sPP, and an ultrathin film of highly regular iPP are

110


Fig. 7.5. (A)–(B) Height images of the top surface of ethylene-octene copolymer (ρ = 0.86g/cm3 ) sample prepared by molding between mica sheets. The images were obtained in light and hard tapping on the same area of 760 × 760 nm2 .

seen in AFM images in Figs. 7.6A–F. In both cases, one can notice single crystallike objects surrounded by smaller crystalline entities. Rectangular platelets of sPP exhibit well-defined sectors reflecting different orientations of the crystalline lattice [32]. Numerous fibrillar structures extending from the crystal edges (Fig. 7.6A–B) are believed to be due to growth in conditions of material depletion. The highresolution image in Fig. 7.6C presents a part of the crystal, where the surface is covered with numerous grains. The straight fibrils grown from the crystal edge also reveal grains and tiny overgrown fibrils oriented in a perpendicular direction, which is most likely the result of the homoepitaxy of polypropylene. AFM images of an ultrathin iPP film (Figs. 7.6D–E) show single crystals of slightly different habit than those of sPP [33]. Nevertheless, they also present different sectors and grainy surface morphology (Fig. 7.6F). Stacks of edge-on standing lamellae and web-like patches formed by tiny fibrils are seen between the single crystals. A more detailed characterization of these morphologic features will be given elsewhere [33]. Additional information on the semicrystalline structure and thermal transitions of LDPE and iPP can be obtained from AFM imaging at elevated temperatures (Fig. 7.7A–H). The first series of images shows changes in morphology of an ultrathin film of LDPE recorded during a step-like heating of this sample. Initial spherulitic morphology of the crystalline LDPE layer has been formed by dipping the substrate into a hot polymer solution [34]. It can be seen that heating of the layer to 80 ◦ C does not bring any substantial modifications to the morphology. By contrast, further heating to 90–100 ◦ C induces drastic changes: at 100 ◦ C, instead of spherulitic structures linear lamellar aggregates having several microns in length and up to 1 micron in width grew in the same sur-

7


111

Fig. 7.6. (A)–(B) Height and phase images (12 × 12 µm2 ) of the surface of fractionated syndiotactic polypropylene. (C) Phase image of a part of the area shown in (A)–(B), 2 µm × 2 µm scan. (D)–(E) Height and phase images of the surface of a spin-cast film of isotactic polypropylene, image size is 20 × 20 µm2 . (F) Height image (3 × 3 µm2 ) presenting a detailed view of the region shown in (D)–(E).

face region. Heating to 115 ◦ C, which is in the main melting region of the bulk LDPE sample, leads to a featureless surface indicative of complete melting of the layer. These observations present direct evidence for recrystallization processes occurring during heating of a LDPE layer. AFM visualization of the structural transformation during recrystallization helps understanding changes in polymer properties upon annealing at these temperatures. Melting of an ultrathin film of iPP demonstrates another kind of structural change in the pre-melting temperature region (Fig. 7.7E–H) [33]. The network of thin lamellar and fibrillar structures visible in the first image of the series melts almost completely at 120 ◦ C (Fig. 7.7F), whereas melting of single crystals of iPP becomes noticeable at 150 ◦ C (Fig. 7.7G). Finally, only small remnants of the initial crystalline structure can be found in the image at 170 ◦ C (Fig. 7.7H). However, crystallization of the molten sample at 160 ◦ C restores the initial morphology consisting of crystalline platelets embedded in a network of tiny lamellae.

112


Fig. 7.7. (A)–(D) Height images (10×10 µm2 ) of ultrathin film of LDPE on Si substrate at T = 25, 80, 100 and 115 ◦ C, respectively. (E)–(H) Height images of ultrathin film of isotactic polypropylene on Si substrate at T = 25, 120, 150 and 170 ◦ C, respectively. Image size is 20 × 20 µm2 .

7.3.1.3 Blends of Poly(ε-caprolactone) and Poly(Vinyl Chloride), PCL/PVC Binary semicrystalline/amorphous blends of PCL and PVC extensively studied in the past [35] present a difficult object for SAXS studies due to their linear crystallinity (ϕc,lin ≡ LLBc ) varying around 0.5 and the electron density contrast becoming faint at certain extents of interlamellar inclusion of PVC. Therefore, high temperature AFM is probably the only technique offering the possibility of in-situ crystallization studies of these systems at the lamellar scale [36]. Samples for AFM studies [37] were prepared by casting blend solutions in THF on freshly cleaved mica to obtain films of approx. 10 µm thickness. The reproducibility of resulting semicrystalline morphologies (e.g., the primary nucleation density) was improved by applying the self-seeding technique. Typical topography and phase images corresponding to the crystallization of PCL/PVC 75/25 wt/wt blend at 40 ◦ C are shown in Fig. 7.8. The whole image sequence (not shown here) corresponding to about 1 hour melt-crystallization was recorded at the same surface spot. At the selected temperature linear growth of PCL crystals occurring during the primary stage of crystallization (i.e., radial propagation of the crystallization front) is too rapid to be observed with AFM. By contrast, the processes of secondary crystallization operating at a slower pace inside the already grown spherulites could be examined with some scrutiny. Since the crystalline lamellae are better defined in the AFM phase images, the latter were used to compute the degree of crystallinity by volume and the morphological parameters of the semicrystalline structure in direct space. The phase images were firstly converted to binary format by choosing a threshold value of the phase signal above which pixels were attributed to

7


113

Fig. 7.8. AFM topography and phase images recorded during isothermal meltcrystallization of a PCL/PVC 75/25 wt./wt blend at 40 ◦ C. Elapsed times are 0 (A,B), 541 s (C,D), and 2931 s (E,F). All images of this sequence were successively recorded on the same surface area: The film defect noticeable in the upper right part of the images was used as a marker. The second image of the series (C,D) roughly corresponds to completion of the primary crystallization stage at the image location. The full gray-scale is 12 nm and 16◦ for topography and phase images, respectively.


Crystallinity (vol/vol)

114

0.4

0.2 Secondary Crystallization Stage

0.0

0

2000 Time (s)

Fig. 7.9. Crystallinity by volume computed from the AFM phase images corresponding to the melt-crystallization of PCL/PVC 75/25 wt./wt. blend at 40 ◦ C.

the crystalline phase. This threshold value was computed by minimizing contour line fuzziness [38], and crystallinity by volume was then calculated as the fraction of pixels above the threshold. The time evolution of this parameter is shown in Fig. 7.9. It can be seen that the first rapid increase of crystallinity is followed by a much slower growth, which is, however, still detectable after about 1 hour of crystallization. The evolution of crystal thickness was computed by analyzing the perimeter (P ) and surface (S) of bright particles (crystalline lamellae) in binary images. The value of Lc was calculated as 2S/P , which gives a good estimate of crystal thickness for sufficiently elongated objects with a high contour persistence length. A small fraction of more roundish particles with Pcirc ≤ 1+ε πε , P2 and ε – a small value, were excluded from consideration in where Pcirc ≡ 4πS c order to keep a certain precision in the thickness determination ( ∆L Lc ≤ ε). The output of this analysis was a surface-weighted lamellar thickness distribution as a function of crystallization time (cf. Fig. 7.10). The results presented in Fig. 7.10 indicate that a slight thickening of PCL crystals occurs during the secondary crystallization stage. The results of the direct-space analysis can be completed by analysis of the same images in reciprocal space, which is performed similarly to the classical treatment of SAXS curves. In the first step, the two-dimensional power spectral density function (P2 (s)) was computed from AFM images (u(r)) up to the critical, or Nyquist, frequency depending upon the experimental sampling interval as: 2 1 P2 (s) ≡ u(r)W (r) exp (2πis · r) d2 r , (7.1) A A

where A denotes the image area, W (r) window function [39] and s the reciprocal space vector. The P2 (s) function was then transformed into the one-dimensional

7


115

Apparent Lc (nm)

12

10

8

6 0

2000

Crystallization time (s)

Fig. 7.10. Surface-weighted distribution of lamellar thickness computed from the AFM phase images measured during the melt-crystallization of PCL/PVC 75/25 wt./wt. blend at 40 ◦ C. The error bars correspond to the standard deviation of the crystal thickness distribution.

PSD (P1 (s)) according to: P1 (s) = (2πs)−1 P2 (s )δ(|s | − s)ds ,

(7.2)

and, finally, the one-dimensional correlation function was computed as a real part of the Fourier transform of P1 (s):    ∞  γ1 (l) = Re 2π P1 (s)s exp(2πisl) exp(4π 2 σ 2 s2 )ds , (7.3)   0

where the P1 (s) function was preliminarily corrected for the presence of sigmoidal gradient crystal/amorphous transition layers having thickness σ [40]. Calculation of the morphological parameters of the semicrystalline structure was performed with a standard approximate relationship [41]: ϕc,lin (1 − ϕc,lin ) = r0 /LB .

(7.4)

In Eq. (7.4), r0 stands for the first intercept with the abscissa of the tangent to the linear part of the correlation function in the self-correlation triangle. The lamellar thickness Lc is found from Eq. (7.4) as ϕc,lin LB , where LB is determined from the location of the first subsidiary maximum of γ1 (l). The resulting correlation functions and the morphological parameters of the semicrystalline structure of the blend are given in Fig. 7.11.

116


( )

Apparent distance (nm)

γ1(l) (deg2)

4

2

0 0

20

40

nm

20

LB

Lc 10 σ

0 0

2000 Time (s)

Fig. 7.11. One-dimensional SAXS-type correlation functions and the morphological parameters of the semicrystalline structure computed from the AFM images given in Fig. 7.8. Open symbols correspond to topography and filled to phase images.

It can be seen that the evolution of lamellar thickness determined by the reciprocal space analysis agrees with the results of the direct space analysis reported before, i.e., the crystal thickness shows a monotonic increasing trend. By contrast, the long period probably decreases slightly with time, which is a rather typical observation for isothermal polymer crystallization. In our case, this slight decrease can be related to the in-filling growth occurring between the primary lamellae formed during the rapid stage of crystallization. Examination of the images shows, however, that insertion takes place only in the amorphous pockets that are large enough to accommodate growth of new crystals, and, therefore, the total variation of LB remains very limited. 7.3.2

Semi-rigid Chain Polymers

While most of AFM work to date has been performed on flexible chain polymers, some important questions related to semicrystalline morphology of semi-rigid chain polymers such as poly(ether-ether ketone), PEEK, and various aromatic polyesters remain open. Generally, despite the fact that the structure of these polymers has been in focus of numerous studies, very little is still known of their lamellar organization. This can be explained by the fact that, on the one hand, these polymers typically present difficult objects for TEM studies [42], and, on the other hand, SAXS data analysis could be ambiguous leading in some cases to very different micro-structural models of these polymers. For example, the uncertainty in Lc as determined from the SAXS correlation function analysis could be as high as 100%. Such uncertainty clearly expresses the need, in studies

7


117

of the semicrystalline structure of these polymers, for new approaches that would simplify the interpretation of SAXS data. In this section, the problems of SAXS data analysis common to the majority of semirigid chain polymers will be briefly reviewed, and the role of AFM in clarifying these controversial issues will be critically analyzed. 7.3.2.1 Poly(Ether Ether Ketone), PEEK, and Its Blends with Poly (Ether Imide), PEI PEEK is a typical member of the family of semirigid chain polymers characterized by a fully para-aromatic backbone structure. It is noted for high performance properties, among others excellent thermal and chemical stability, absence of polymorphic transitions, and an extremely large temperature interval of crystallization. The lamellar arrangement of PEEK has been a subject of heated debate due to a difference between authors in the interpretation of the SAXS results for this polymer [43,44,45]. Briefly, the proposed microstructural models can be divided into two groups. The first model is essentially a three-phase model, wherein thick lamellar crystals (about 70–120 ˚ A) separated by thinner amorphous interlayers (about 30–40 ˚ A) pack into stacks of finite size, these stacks being separated by large purely amorphous PEEK regions. The large amorphous gaps represent a significant fraction of the polymer (up to 50% according to this model). The second model is essentially a two-phase model, wherein the sample is homogeneously filled with stacks of thin lamellae alternating with thicker amorphous interlayers. Different arguments in favor of one or the other model were discussed in some detail elsewhere [45]. In short, the main reason for the difference of opinion between authors is due to the Babinet reciprocity theorem, which states that, in a two-phase system the electron densities of the phases may be interchanged, without affecting the scattering curve or, consequently, the correlation function [46]. In other words, the solution that one finds for example from the classical one-dimensional correlation function analysis is not unique, as the attribution of the two distances obtained from this analysis to the crystalline and amorphous layer thicknesses (i.e., Lc and La ) can be inverted. Thus, clearly some additional structural information would be required to choose between the two models described above, and simple accumulation of SAXS data treated with the help of the correlation function will not provide any new insight. In one of the attempts to solve this problem [47], a direct study of the semicrystalline structure of PEEK was carried out with AFM. In this case, imaging was performed on miscible blends of PEEK with poly(ether imide) (PEI), an amorphous polymer fully miscible with PEEK in the amorphous state. The use of PEI as an inert macromolecular diluent made it possible to open up the dense semicrystalline structure of PEEK and visualize the details of the subspherulitic organization in the course of its evolution (cf. Fig. 7.12). Since the heating stage accessory for AFM was not readily available at the time of the measurements, the imaging had to be performed ex situ. In this experiment, the sample was quenched to room temperature after different times of crystallization at Tc or upon heating to different annealing temperatures Ta (Ta > Tc ) and imaged

118


at room temperature. To achieve the visualization of the time- or temperatureevolution of a single crystalline entity (e.g., single spherulite or lamellar stack), a special nanopositioning technique was employed to repeatedly retrieve the same spot on the sample surface with a precision of several tens of nanometers. The results of this rather tedious approach show (Fig. 7.12) that AFM can indeed provide information on the evolution of single elements of the semicrystalline structure of PEEK. More specifically, the obtained image sequences gave access to the growth rates of individual lamellar bundles [47], which is impossible with Polarized Optical Microscopy, and to the spatial localization of the reorganization process occurring on heating. However, the imaging conditions employed in the experiments did not allow to obtain lamellar resolution, which somewhat limited the interest of such studies for structural analysis and, consequently, confined the potential of AFM in this field to only qualitative comparison with X-ray results. It became clear that high-resolution in situ AFM imaging was needed to address the issues of lamellar organization of semirigid chain polymers. Such morphological AFM work has still to be done on PEEK [48]. 7.3.2.2

Poly(Ethylene Terephthalate), PET

PET is a high-performance aromatic polyester currently used for many industrial applications. The first studies on the semicrystalline structure of PET were carried out more than 40 years ago [49]. Since then, a large amount of data has been obtained on the structure and crystallization behavior of this polymer using different experimental techniques ranging from X-ray and electron diffraction to small-angle light scattering, and from differential scanning calorimetry to IR/Raman spectroscopy. Generally, PET is a low crystallinity polymer sharing many similarities with PEEK, introduced about two decades later. In particular, quite similarly to the situation with PEEK, the information available on the microstructure of PET is very scarce and, therefore, the same problem with the SAXS data analysis (i.e., the possibility of interchanging the distances Lc and La ) is typically encountered [50]. It is worth noting that the problem with SAXS data analysis not only has direct consequences upon the choice of the structural model for lamellar ar-

Fig. 7.12. Square 25 µm2 topographical AFM images corresponding to the growth of a single PEEK spherulite in a PEEK/PEI 70/30 wt./wt. blend cold-crystallized at Tc = 180 ◦ C. Elapsed times correspond to 4 (A), 8 (B), 12 (C) and 60 min (D). Imaging in contact mode was performed ex situ (see text for more details).

7


119

Fig. 7.13. Schematic diagrams of evolution of lamellar microstructure during isothermal crystallization: Dual lamellar population model (cf. [43] and references therein) containing separate stacks of thick and thin lamellae (A) and mixed stacks comprising both types of lamellae (B). In this model linear crystallinity in the stack is significantly higher than bulk crystallinity. To account for this difference, large amorphous gaps are supposed to exist in the structure.

rangement but also affects the interpretation of the multiple melting behavior well-documented for PEEK, PET and other semirigid chain polymers [44,51]. Thus, the proponents of the first structural model (see previous section for details) explain the decrease of long period during isothermal annealing either by the formation of separate stacks of thinner crystals (Fig. 7.13A), or by insertion of thinner, more defective, crystals in-between the primary (dominant) ones (Fig. 7.13B). In both cases, the low-temperature endothermic peaks present in DSC scans of the annealed polymer would correspond to melting of these later grown, ill-formed crystals. By contrast, the proponents of the second model (i.e., homogeneous, or infinite stack model) attribute multiple melting behavior to the melting-reorganization process in which the majority of crystals undergo a continuous transformation, i.e., melting followed by immediate recrystallization, to form thicker, more perfect and thermodynamically stable crystals. Since the technique of DSC is unable, despite its high sensitivity, to provide any information regarding the spatial localization of the melting processes, further studies are required to check the validity of each of the crystallization models. The development of AFM made it possible to visualize the details of PET semicrystalline morphology [52]. However, imaging of PET in tapping mode at ambient temperature normally does not yield sufficient phase contrast between the crystalline and amorphous regions. To improve the contrast, one would need to use tapping mode at temperatures higher than the glass transition (Tg ) of the amorphous interlamellar regions, at which the difference in the mechanical properties of both phases becomes much more pronounced. In our recent work [53], the melt-crystallization of PET was monitored in situ with high temperature AFM. Typical images of crystallization kinetics acquired at 233 ◦ C are shown in Fig. 7.14. It can be seen that the PET lamellae display a marked tendency to form stacks. The arrows in the figure point to locations where crystallization mainly proceeds by growth of crystals parallel to the borders of the already existing stacks, which gives rise to the overall increase of the number of crystals per stack. This mechanism termed “stack thickening” seems to play an important role at

120


Fig. 7.14. AFM phase images (1 µm2 ) recorded in hard tapping during crystallization of PET from the melt at 233 ◦ C; elapsed times are 72 (A), 80 (B), 96 (C), and 112 min (D). Thin white stripes correspond to the crystalline lamellae in almost edgeon orientation. The arrows help the reader to identify the same lamellar stacks and locations on the sample surface where crystallization mainly proceeds via the stack thickening mechanism.

the secondary crystallization stage. By contrast, only very limited occurrence of insertion growth was detected upon thorough examination of the images. In addition, these rare insertion events were found to occur only in those amorphous zones which were much larger than the “normal” interlamellar regions, which is very different from the scheme depicted in Fig. 7.13B. The time-evolution of the semicrystalline structure of PET was studied by direct space AFM image analysis [53]. The average surface-weighted crystal thickness shown as a function of crystallization time in Fig. 7.15 does not reveal any evolution, at least in the time scale of the experiment. This fact is at variance with the first micro-structural model described earlier (cf. Fig. 7.13). It should be mentioned, however, that the fraction of the subsidiary crystals is a priori unknown. Thus, to exclude the possibility that this lamellar population has been overlooked in the treatment due to insufficiency of the image statistics, further analyses were performed. To detect this possibly missing lamellar population, it was decided, first, to increase the number of analyzed crystals (i.e., perform the same analysis on several images) and, second, to inspect the semicrystalline morphologies formed after long-time crystallizations. Typical images of the PET morphology formed after completion of crystallization are presented in Fig. 7.16. It is clear that no amorphous gaps of large size (which was

7


121

Apparent Lc , nm

15

10

5

0 80 100 120 Elapsed Time, min

Fig. 7.15. Evolution of the average crystal thickness computed from the AFM images given in Fig. 7.14. The error bars indicate the standard deviations of the distributions.

earlier estimated to be on the order of 200 nm [43]) could be visualized in the images: the semicrystalline structure appears to be completely space-filling. This observation necessarily implies that the linear crystallinity determined by SAXS should be close to the bulk crystallinity of the sample. In addition, the apparent crystal thickness distribution at the end of crystallization remains monomodal as it was during the crystallization kinetics reported in Fig. 7.14. The average crystal thickness decreases with temperature (cf. Fig. 7.16C), which is in agreement with kinetic theories of crystallization. Thus, the hypothesis of secondary crystallization of PET producing more imperfect (thinner) crystals is not supported in our experiments. The periodicity of lamellar stacking determined from the images obtained at Tc = 233 ◦ C was quantitatively compared to the SAXS results obtained on the same sample (cf. Fig. 7.16D). The AFM and SAXS values of LB found from the corresponding correlation functions γ1 (r) are rather close (19.2 and 15.5 nm, respectively), which ensures that the morphological parameters of the semicrystalline structure determined by these two very different techniques are comparable. The AFM linear crystallinity calculated using the results of the direct and reciprocal space analysis of the images is given as ϕc,lin = 10/19.2 ∼ 0.5 (cf. Figs. 7.15 and 7.16D); the corresponding SAXS values equal either 0.4 or 0.6, depending upon the attribution of the two distances to the sizes of one or the other phase. The value of the AFM linear crystallinity is thus intermediate between the two possible SAXS values, which would formally render the choice difficult. However, it is clear that the bulk of high temperature AFM results can be reconciled only with the choice of the SAXS crystal thickness as the smallest distance, i.e., Lc < La . Indeed, in the opposite case, the fraction of large nonc ∝ 0.33–0.42, where crystallized inter-stack regions amounting to about 1 − ϕϕc,lin

122


Fig. 7.16. (A, B) Semicrystalline morphology of PET after completion of meltcrystallization at 233 and 220 ◦ C, respectively. AFM phase images were obtained at the temperature of crystallization. (C, D) Quantitative analysis of semicrystalline morphology of PET. (C) Distributions of the apparent lamellar thickness corresponding to images (A) (lower histogram) and (B) (upper histogram). (D) One-dimensional correlation functions computed from the AFM image shown in (A) and from SAXS data obtained on the same sample at room temperature; curves are vertically offset for clarity.

ϕc is bulk crystallinity of the sample, should have been distinctly observed in the images of a completely crystallized sample of PET. Another argument in favor of the homogeneous model is that the AFM value of ϕc,lin could only be higher than the SAXS linear crystallinity because of the previously mentioned effect of dilation of surface features by the tip geometry. The dilation increases the apparent thickness of crystals slightly protruding the sample surface, without

7


123

Fig. 7.17. Model of the stack thickening process operating during the secondary stage of the melt-crystallization of PET. Visualized with high temperature AFM, this process consists in progressive increase of the size of lamellar stacks containing thin crystals separated by thick interlamellar amorphous regions.

modifying the intercrystalline distance, and thereby increases the apparent linear crystallinity. To sum up, a direct assessment of melt-crystallization of PET by in situ AFM shows that the evolution of the semicrystalline structure can be mainly accounted for by the stack thickening process depicted in Fig. 7.17. In this process, new crystals grow parallel to the borders of the already existing stacks described by the homogeneous micro-structural model, i.e., containing thin crystals separated by thicker amorphous interlayers. It is important to note that the stack thickening model can successfully account for a decrease of the SAXS long period upon annealing [49] if lamellar stacks contain second-order defects in the structure, or paracrystallinity. This effect is described in some detail elsewhere [54]. 7.3.2.3

Poly(Trimethylene Terephthalate), PTT

PTT is a recently commercialized aromatic polyester, with a chemical structure intermediate between that of PET and poly(butylene terephthalate), PBT. This polymer has already been the subject of several publications focusing on its crystal structure, as well as crystallization and melting behavior [55,56]. A detailed study of the organization of PTT at the nanometer scale and its evolution during crystallization has recently been carried out by high temperature AFM and time-resolved SAXS [57]. The AFM images given in Fig. 7.18 are typical of crystallization kinetics observed at 210 ◦ C. They show the propagation of the crystallization front in the form of large leaf-like lamellar crystals. Interestingly enough, in these images one can note the appearance of a spherulitic band formed by the constituent S-shaped lamellar stacks marked with white arrows. Generally, melt-crystallized PTT displays banded spherulitic texture, the spacing of which was found to vary with crystallization conditions [56]. Although not expected at such high crystallization temperature [56], pronounced banding of large PTT spherulites grown at 210 ◦ C was observed optically and inspected at

124


Fig. 7.18. Melt-crystallization of PTT, as visualized with in situ high temperature AFM at 210 ◦ C. The white arrows point at S-shaped lamellar stacks forming a spherulitic band, which extends from the top center toward the left bottom of the images. Elapsed times: (A)–0, (B)–8 min.

Fig. 7.19. High temperature AFM images of the semicrystalline structure of PTT: crystallization and imaging at 210 ◦ C. Structure of a spherulitic band, as observed at the micrometer (A) and sub-micrometer (B) scales. The arrow indicates the onset of a concerted lamellar twist involving many crystals.

different spatial scales with AFM (Fig. 7.19). At the scale of several microns only a discontinuous morphological change related to spherulitic bands (Fig. 7.19A) can be observed, whereas images at the sub-micrometer scale reveal a neatly resolved concerted lamellar twist involving many lamellae (Fig. 7.19B) [58]. The onset of the twist can be identified in the images by the curvature of otherwise straight lamellae pointed with an arrow in Fig. 7.19B. However, it is technically difficult to follow the propagation of individual lamellae throughout the entire twist period because AFM provides only a two-dimensional projection of the lamellar structure. Therefore, the absence of certain elements of the structure, e.g., parts of lamellae buried underneath the surface, complicates the reconstitution of the three-dimensional structure.

7


125

The presence of a regular lamellar twist (rhythmic growth) and the dominance of lamellar stacks containing a large number of crystals make the assessment of the semicrystalline structure of PTT with AFM easier than that of other aromatic polyesters such as PET or PBT. The sub-micrometer AFM phase images of PTT obtained above the Tg of interlamellar amorphous regions normally reveal well defined lamellar crystals (cf. Fig. 7.19B). The parameters of the semicrystalline structure can then be evaluated for the regions where the crystals are grown close to edge-on orientation. Typical distributions of the crystal thickness and long period for the semicrystalline morphology of PTT formed at 210 ◦ C are given in Fig. 7.20. 3

A

15 10 5 0 0

20x10 Frequency (a.u.)

Frequency (a.u.)

20x10

10

20

30

40

nm

3

B

15 10 5 0 0

10

20

30

40

nm

Fig. 7.20. Distributions of crystal thickness (A) and long period (B) corresponding to the semicrystalline morphology of PTT formed at 210 ◦ C. The distribution functions were calculated from the direct space analysis of AFM images using home-built object recognition routines [57].

The comparison of the distribution functions shows that the width of the long period distribution is higher than that of crystals, which is logical since the former is the convolution of the latter with the corresponding distribution of amorphous interlamellar regions. The quantitative analysis of the distributions shows that crystals are slightly thinner than the interlamellar amorphous regions, which is similar to the results on melt-crystallized PET reported earlier. The position and shape of these distributions will be further directly compared [57] with those extracted from the synchrotron SAXS data analyzed with the generalized one-dimensional paracrystalline model of Hosemann [59,54]. 7.3.2.4

Syndiotactic Polystyrene, sPS

A broad use of metallocene catalysts led to synthesis of a variety of polymers with well-controlled macromolecular structures. In addition to polyolefins, other polymers such as syndiotactic polystyrene are being produced in large quantities. sPS exists in several crystal modifications, which can be recognized not only by their X-ray diffraction patterns but also by the resulting lamellar morphologies [60,61]. Similar to the case of PCL/PVC blends, a relatively small difference in

126


electron density between amorphous and crystalline regions of sPS limit the use of X-ray scattering techniques for the analysis of its semicrystalline structure [62]. Therefore, AFM can be particularly useful for morphological characterization of sPS in the sub-micrometer range. A series of images corresponding to the α- and β-forms of melt-crystallized sPS are given in Fig. 7.21 [63]. At a large scale, randomly oriented fibrillar entities represent the semicrystalline morphology of α-sPS, and it is best seen in the 5-micron image in Fig. 7.21A. It was specifically checked that this morphology is also representative for the bulk of the α-sPS samples. Quite loose packing is seen not only for the fibrils at the scale of microns but also for lamellar arrangement at the sub-micron scale. In some locations stacks of edge-standing lamellae can be found, which is exemplified in Fig. 7.21B. The calculation of the SAXS-type interface distribution function for this image (Fig. 7.21C) gives 8.5 nm width for both crystalline and amorphous parts of the structure. By contrast, the β-form of sPS is characterized by welldefined spherulites (Fig. 7.21D) up to 100 micron in size. A stack of the edge-on

Fig. 7.21. AFM images of melt-crystallized syndiotactic polystyrene. (A)– (B) Semicrystalline morphology of the α-form of sPS showing fibrillar and lamellar organization at the scale of 5 and 1 µm, respectively. (C) SAXS-type interface distribution function computed from the image in (B). (D)–(E) Semicrystalline morphology of the β-form of sPS presenting large spherulites with tightly packed lamellar stacks, image size 100 µm (D) and 1 µm (E). (F) Interface distribution function corresponding to the image in (E). Images (B) and (E) are recorded at 220 ◦ C.

7


127

standing lamellae (Fig. 7.21E) was chosen to obtain a quantitative estimate of lamellae packing (Fig. 7.21F). It was found that the width of the crystalline and amorphous regions is 15.5 nm and 9.5 nm, respectively. Consequently, linear crystallinity and long period of β-sPS are larger than those of α-sPS.

7.4

Conclusions

The potential of variable temperature AFM in studies of semicrystalline polymers is described. The advantages and drawbacks of this technique with respect to classical optical and electron microscopies are briefly reviewed. It is shown how the advantageous characteristics of AFM such as its non-destructive character, high precision height measurements and possibility to perform in situ studies at different temperatures can bring new insights into the structure and thermal behavior of semicrystalline polymers. Several examples including flexible and semi-rigid chain polymers illustrate how AFM is complementary to conventional scattering techniques providing quantitative direct-space structural information that can be used to interpret the results of scattering experiments.

Acknowledgements The authors appreciate the contribution of Natalya A. Yerina (Digital Instruments/Veeco Metrology Group), who was involved in AFM studies described in this work. D.A.I. is grateful to M. Buscemi (Shell Coordination Center, Belgium) for providing a sample of PTT. The authors would like to thank Dr N. Heymans (Université Libre de Bruxelles, Belgium) for reviewing the manuscript.

References 1. Binnig, G.; Rohrer, H.; Gerber, Ch.; Weibel, E. Phys. Rev. Lett. 1982, 49, 57. 2. Binnig, G.; Quate, C.; Gerber, Ch. Phys. Rev. Lett. 1986, 56, 930. 3. Aleksander, S.; Hellemans, L.; Marti, O.; Schneir, J.; Elings, V., Hansma, P. K.; Longmire, M.; Gurley, J. J. Appl. Phys. 1989, 65, 164. 4. Zhong, Q.; Innis, D.; Kjoller, K.; Elings, V. Surf. Sci. Lett. 1993, 290, L688. 5. “Exploring the High-Temperature AFM and Its Use for Studies of Polymers” Ivanov, D.A.; Daniels, R.; Magonov, S. Application Note published by Digital Instruments/Veeco Metrology Group (2001), pp. 1–12. Available on line at URL: http://di.com/APPNotes PDFs/AN45%20HeatingStage.pdf 6. Most experimental work was done with custom-made accessories whose temperature range was limited below 100–130 ◦ C. 7. Magonov, S. N.; Whangbo, M.-H. “Surface Analysis with Scanning Tunneling and Atomic Force Microscopy”, VCH, Weinheim, 1996; Scanning Probe Microscopy of Polymers, Ratner, B.; Tsukruk, V.V., Eds., ACS Symposium Series, ACS Washington, DC, 1998, 694. 8. Magonov, S. N. “AFM in Analysis of Polymers” Encyclopedia of Analytical Chemistry, (R. A. Meyers, Ed.), pp. 7432–7491, John Willey & Sons Ltd, Chichester, 2000.

128


9. Jandt, K.D. Mat. Sci. Eng., 1998, R21, 221; Morris, V. J.; Kirby, A. R.; Gunning, A. P. “Atomic force microscopy for biologists”, World Scientific Publishing, Singapore, 1999. 10. Vertical displacements of a probe or a sample are obtained with a feed-back controlled Z piezo-actuator, whereas independent lateral XY-motion is performed by another piezo-actuator. 11. Magonov, S. N.; Elings, V.; Whangbo, M.-H. Surf. Sci. Lett. 1997, 375, L385. 12. Cleveland, J. P.; Anczykowski, B.; Schmid, A. E.; Elings, V. B. Appl. Phys. Lett. 1998, 72, 2613. 13. In fact, the height difference between surface locations with different density is largest when the set-point amplitude is lowered below 20% of its initial value. Yet imaging at such amplitudes is performed practically in quasi-contact mode and sample damage could be pronounced. 14. Magonov, S. N.; Elings, V.; Cleveland, J.; Denley, D.; Whangbo, M.-H. Surface Science 1997, 389, 201. 15. Ponomarenko, S. A.; Boiko, N. I.; Zhu, H.-M.; Agina, E. V.; Shibaev, V. P.; Magonov, S. N. Polymer Science 2001, 43, 419. 16. Prokhorova, S.A.; Sheiko, S.S.; Ahn, C.-H.; Percec, V.; M¨ oller, M. Macromolecules 1999, 32, 2653. 17. Percec, V.; Holerca, M. N.; Magonov S. N.; Yeardley, D. J. P.; Ungar, G.; Duan, H.; Hudson, S. D. Biomacromolecules 2001, 2, 706. 18. Bassett, D. C., Principles of Polymer Morphology, Cambridge University Press, New York, 1981. 19. Geil, P. H. Polymer Single Crystals, 1963, John Willey & Sons. 20. Patil, R.; Kim, S.-J.; Smith, E.; Reneker, D. H.; Weisenhorn, A. L. Polym. Comm. 1990, 31, 455. 21. Patil, R.; Reneker, D. H. Polymer 1994, 35, 1909; Stocker, W.; Magonov, S. N.; Cantow, H.-J.; Wittmann, J.-C.; Lotz, B. Macromolecules 1993, 26, 5915; Nie, H.-Y.; Motomatsu, M.; Mizutani, W.; Tokumoto, H. Polymer 1996, 36, 183. 22. Tian, M.; Loos, J. J. Polym. Sci. Phys. 2001, 39, 763. 23. Recent attempts are directed towards modeling of this particular solid state transition, as well as that of the polymer crystal growth in general using computer simulations: Welch, P.; Muthukumar, M. Physical Review Letters 2001, 87, 218302; Sommer, J.-U.; Reiter, G. Europhys. Lett. 2001, 56, 755; Anderson, K.L.; GoldbeckWood, G. Polymer 2000, 41, 8849. 24. Ungar, G.; Zeng, K.B. Chem. Rev. 2001, 101, 4157; Hobbs, J.K.; Hill, M.J.; Barham, P.J. Polymer 2000, 41, 8761; Hobbs, J.K.; Hill, M.J.; Barham, P.J. Polymer 2001, 42, 2167. 25. Although the stacking periodicity of these crystals (19 nm) is approximately equal to that in the bulk, their habit is closer to fibrils since the width and thickness of these crystals are comparable. 26. Magonov, S. N.; Yerina, N. A.; Ungar, G.; Reneker, D.; Ivanov, D., in preparation. 27. The thickness of the microtomed slice is supposed to be constant. 28. Magonov, S. N.; Godovsky, Yu. K. Amer. Lab. 1999, 31 , 52. 29. Kanig, G. Colloid Polym. Sci. 1991, 269, 1118. 30. Michler, G. H. Kunstoff-Mikromechanik, 1992, p. 187, Carl Hanser Verlag. 31. Hugel, T.; Strobl, G.; Thomann, R. Acta Polym. 1999, 50, 214. 32. Zhu, W.; Cheng, S. Z. D.; Putthanarat, S.; Eby, R. K.; Reneker, D. H.; Lotz, B.; Magonov, S.; Hsieh, E. T.; R. G. Geerts; Palackal, S. J.; Hawley, G. R.; Welch, M. B. Macromolecules 2000, 33 , 6861.

7


129

33. Yerina, N. A.; Magonov, S. N.; Jaaskalainen, P., in preparation. 34. Godovsky, Yu. K.; Magonov, S. N. Langmuir 2000, 16, 3549; Godovsky, Yu. K.; Magonov, S. N. Polymer Science 2001, 43, 1035. 35. Khambatta, F. B.; et al. J. Polym. Sci. Polym. Phys. 1976, 14, 1391; Russell, T. P.; Stein, R. S. J. Polym. Sci. Polym. Symp. 1983, 21, 999; Chen, H.-L.; Li, L.-J.; Lin, T.-L. Macromolecules 1998, 31, 2255. 36. Basire, C.; Ivanov, D.A. Physical Review Letters 2000, 85, 5587; Basire, C.; Ivanov, D.A. (unpublished data). 37. The samples of PCL and PVC were obtained from SOLVAY S.A. (Solvic grades CAPA 650 and 258RD). The molecular weights Mn and polydispersity indices, as determined by gel permeation chromatography (GPC), are 85 400, 1.55 and 34 600, 1.87, respectively. 38. Russ, J. C. The Image Processing Handbook, CRC Press, Boca Raton, FL, 1998. 39. Press, W. H.; et al. Numerical Recipes in C, The Art of Scientific Computing (Plenum Press, New York 1988). 40. Koberstein, J. T.; Morra, B.; Stein, R. S. J. Appl. Crystallogr. 1980, 13, 34. 41. Strobl, G.R.; Schneider, M. J. Polym Sci: Part B 1980,18,1343. 42. Watkins, N.C.; Hansen, D. Textile Res. J. 1968, 32(4), 388; Ivanov, D.A.; Pop, T.; Yoon, D.Y.; Jonas, A.M. Polym. Mater. Sci. Eng. 1999, 81, 335; Ivanov, D.A.; Pop, T.; Yoon, D.Y.; Jonas, A.M. (Macromolecules, submitted) 43. Verma, R. K.; Hsiao, B. S. Trends Polym. Sci. 1996, 4, 312. 44. Blundell, D. J. Polymer 1987, 28, 2248. 45. Jonas, A.M.; Ivanov, D.A.; Yoon, D.Y. Macromolecules, 1998, 31, 5352; Ivanov, D.A.; Jonas, A.M.; Legras, R. Polymer, 2000, 41, 3719. 46. Balta-Calleja, F.J.; Vonk, C.G. “X-ray scattering of synthetic polymers”, Amsterdam: Elsevier, 1989. 47. Ivanov, D.A.; Jonas, A.M. Macromolecules, 1999, 32, 1582; Ivanov, D.A.; Nysten, B.; Jonas, A.M. Polymer 1999, 40, 5899. 48. A systematic AFM study has been recently started in our laboratory using narrow molecular weight fractions of PEEK synthesized by Dr. J. Roovers. 49. Zachmann, H. G.; Schmidt, G. F. Makromol. Chem. 1962, 52, 23; Elsner, G.; Zachmann, H. G.; Milch, J. R. Makromol.Chem. 1981, 182, 657; Elsner, G.; Koch, M. H. J.; Bordas, J.; Zachmann, H. G. Makromol. Chem. 1981, 182, 1262. 50. Wang, Z. G.; Hsiao, B. S.; Fu, B. X.; Liu, L.; Yeh, F.; Sauer, B. B.; Chang, H.; Schultz, J. M. Polymer 2000, 41, 1791. 51. Lee, Y.; Porter, R. S. Macromolecules 1987, 20, 1336; Lee, Y.; Porter, R. S.; Lin, S. J. Macromolecules 1989, 22, 1756; Holdsworth, P. J.; Turner-Jones, A. Polymer 1971, 12, 195; Medellin-Rodriguez, F. J.; Phillips, P. J.; Lin, J. S.; Campos, R. J. Polym. Sci., Polym. Phys. 1997, 35, 1757. 52. Dinelli, F.; Assender, H. E.; Kirov, K.; Kolosov, O. V. Polymer 2000, 41, 4285. 53. Ivanov, D.A.; Amalou, Z.; Magonov, S.N. Macromolecules 2001, 34, 8944. 54. Ivanov, D.A.; Jonas, A.M. Macromolecules, 1998, 31, 4546. 55. Yang, J.; Sidoti, G.; Liu, J.; Geil, P.H.; Li, C.Y.; Cheng, S.Z.D. Polymer 2001, 42, 7181; Chung, W.T.; Yeh, W.J.; Hong, P.D. J. Appl. Polym. Sci. 2002, 83, 2426. 56. Wang, B.J.; Li, C.Y.; Hanzlicek, J.; Cheng, S.Z.D.; Geil, P.H.; Grebowicz, J.; Ho, R.M. Polymer 2001, 42, 7171. 57. Amalou, Z.; Hocquet, S., Dosiere, M.; Ivanov, D.A. (in preparation). 58. Such a twist has been usually invoked to explain the formation of banded spherulites. 59. Hosemann, R.; Bagchi, S.N. (1962), “ Direct Analysis of Diffraction by Matter ”, North Holland, Amsterdam.

130


60. Greis, O.; Xu, Y.; Asano, T.; Petermann, J., Polymer 1989, 30, 590. 61. Lopez, L. C.; Cieslinski, R. C.; Putzig, C. L.; Wesson, R. D. Polymer 1995, 36, 2331. 62. Barnes, J. D.; McKenna, G. B.; Landes, B. G.; Bubeck, R. A.; Bank, D. Polym. Eng. Sci. 1997, 37, 1480. 63. Yerina, N. A., et al., in preparation.